How Brex is keeping up with AI by embracing the ‘messiness’
In today’s rapidly evolving technological landscape, companies worldwide face a formidable challenge: integrating Artificial Intelligence (AI) tools at a pace that matches their swift development, often clashing with traditional, slow-moving corporate procurement cycles. Corporate credit card innovator Brex found itself at this very crossroads, grappling with the same hurdles as its larger enterprise counterparts. Its response? A radical overhaul of its software procurement strategy to ensure it remains at the forefront of AI adoption.
During the HumanX AI conference in March, James Reggio, CTO of Brex, revealed to TechCrunch the initial struggles. The company’s standard, months-long piloting process for new software proved to be a significant bottleneck in the face of AI’s relentless progress. “In the first year following ChatGPT, when all these new tools were coming on the scene, the procurement process itself would actually run so long that the teams that were asking to procure a tool lost interest in the tool by the time that we actually got through all of the necessary internal controls,” Reggio explained.
This realization prompted Brex to fundamentally rethink its approach. The company initiated a new framework, streamlining data processing agreements and legal validations for AI tools. This innovative system significantly accelerated the vetting process, getting promising tools into the hands of testers much faster than before.
Brex introduced what Reggio terms a “superhuman product-market-fit test.” This unique strategy empowers employees with a crucial role in tool selection. Instead of top-down mandates, the company leverages insights from those actively using the tools, prioritizing solutions that deliver tangible value and drive efficiency within their daily workflows.
A key component of this agile strategy is financial empowerment: Brex provides its engineers with a monthly budget of $50 to license any software tools from an approved list. “By delegating that spending authority to the individuals who are going to be leveraging this, they make the optimal decisions for optimizing their workflows,” Reggio stated. This decentralized approach has yielded interesting results, as the company hasn’t observed a mass convergence on a single popular tool like ‘Cursor,’ validating the decision to facilitate diverse tool trials.
This method has also proven invaluable in identifying broader licensing needs based on accurate usage data from engineers. Brex’s agile procurement has led to a dynamic ecosystem of AI tools. While they currently operate with thousands of AI tools internally, this flexible system also allows them to efficiently cancel and not renew deployments that fail to prove their long-term worth, with approximately five to ten larger deployments having been discontinued.
In essence, Reggio’s advice for enterprises navigating the current AI innovation cycle is to “embrace the messiness.” He stresses the importance of accepting that the journey to finding the right AI tools will be iterative and imperfect, and that’s perfectly acceptable. “Knowing that you’re not going to always make the right decision out of the gate is just like paramount to making sure that you don’t get left behind,” Reggio concluded. He warns against the pitfalls of overthinking and extended evaluation periods, emphasizing that in a landscape as fluid as AI, delaying adoption for perfection can mean being left behind.
